Height profiles of the ionospheric electron density derived using
space-based remote sensing of UV and X-ray emissions, and EISCAT radar data: A ground-truth experiment
A. Aksnes
, J. Stadsnes , N. Østgaard , G. A. Germany
, K. Oksavik
, R. R. Vondrak
, A.
Brekke
, U. P. Løvhaug
Short title: GROUND-TRUTH EXPERIMENT
Department of Physics and Technology, University of Bergen, Norway
Now at the Florida Space Institute, MS-FSI, Kennedy Space Center, FL 32899, U.S.A.
University of Alabama in Huntsville, AL 35899, U.S.A.
Johns Hopkins University/Applied Physics Laboratory, 11100 Johns Hopkins Road Laurel, MD 20723, U.S.A.
NASA/Goddard Space Flight Center, Greenbelt, MD 20771, U.S.A.
University of Tromsø, Department of Physics, Norway
Abstract. In this study we have derived height profiles of the ionospheric electron density
using remote sensing of UV and X-ray emissions from the Polar satellite, and EISCAT radar data. The latter technique gives the most accurate determination of providing a means to ground-truthing the satellite imaging measurements. The UV-emission data are taken from the Ultraviolet Imager (UVI) on Polar, while the X-ray data are measured by the Polar Ionospheric X-ray Imaging Experiment (PIXIE). As UVI yields a far better resolution in time and space than PIXIE, our primary approach involves UVI and EISCAT data. For a substorm event occurring on 24 March 1998, we derive profiles valid in the E-region above 105 km. By comparing with simultaneous values, we find that the two techniques match fairly well in many cases.
The altitude of maximum electron density is usually below 110 km. A few cases reveal
maxima in the upper E-region (130-150 km), indicating a very soft precipitating electron energy spectrum. During such conditions, we observe the largest discrepancies between the and profiles. This may reflect the difficulty of obtaining proper energy characteristics from UV-emissions, when the mean electron energy is less than 2 keV. A recalculation of these values has been performed, requiring that the altitudes of the
maximum must match the altitudes of the maximum. The results reveal a much better agreement between the two data sets, suggesting that UVI is measuring about the same energy flux as EISCAT. Even though the modified values deviate strongly from the old profiles, the effects on the Pedersen conductance, , are insignificant. Also, we find that are within 30 of !" for 15 of 18 cases, suggesting that remote sensing of UV-emissions provide a fairly reliable tool to monitor the Pedersen conductance.
We have investigated a second approach by including PIXIE X-ray data to derive
values valid in the whole E-region and upper D-region. Despite the coarse PIXIE resolution we observe a fairly good match with the profiles. By calculating the Hall and Pedersen conductances, and , we find that the values derived from satellite imaging measurements are within 25 of the EISCAT values for all 4 cases, supporting the space-based remote sensing technique to investigate the ionospheric electrodynamics.
The results presented in this study suggest that the procedures developed to derive
values from the satellite imaging measurements are reliable. We also find that the
and values on average are slightly larger (5 and 13 ) than the values. These discrepancies may be caused by the difference in resolution between the satellite remote sensing data and the radar data, as smoothing of discrete precipitation may result in an overestimation of .
1. Introduction
Space-based remote sensing of visible, ultraviolet (UV) and X-ray emissions provides a powerful tool to monitor the temporal variations of auroral activity on a global scale. In the last 30 years several missions equipped with auroral imagers have been flown in space, yielding valuable information about our near-Earth space. A great advantage with the remote sensing technique is that the particle precipitation morphology may be resolved from the satellite imaging measurements, allowing us to study the time development of ionospheric parameters, like the electron density , on a global scale.
Despite the good insight in ionospheric electrodynamics using the remote sensing technique, we must keep in mind that the procedures mentioned above, of first deriving precipitating particle energy spectra and thereafter , are most complicated. Consequently, validation studies are needed to ensure the reliability of the estimated parameters. The most accurate determination of
is presumably provided by incoherent scatter radars [Robinson and Vondrak, 1994], which clearly suggests that profiles derived from satellite remote sensing data should be compared with using radar measurements. A complication with such ground-truth experiments is the difference in temporal/spatial resolution. Still a number of studies have been performed, supporting various remote sensing techniques. In a study by Robinson et al. [1989] measurements of far-ultraviolet (FUV) emissions by the Dynamic Explorer (DE)-1 satellite were compared with ionospheric conductances derived using Chatanika radar data. A good agreement was found between the variations in FUV emissions and the conductances. Vondrak et al. [1988] derived height profiles of using X-ray data
from the Defense Meteorological Satellite Program (DMSP) F2 satellite. These values were then compared with simultaneously measured using the Chatanika radar, revealing good agreement. A similar conclusion was reached by Robinson et al. [1992] comparing profiles derived using FUV emissions from the Polar Bear satellite, and measured by the Sondrestrom radar.
In 1996 the Polar satellite was launched to study the regions over the poles of the Earth.
The Ultraviolet Imager (UVI) [Torr et al., 1995] and the Polar Ionospheric X-ray Imaging Experiment (PIXIE) [Imhof et al., 1995] onboard Polar measure UV and X-ray emissions respectively. Several studies have estimated the precipitating electron energy spectra using UVI and/or PIXIE measurements [e.g. Brittnacher et al., 1997; Germany et al., 1997; Lummerzheim et al., 1997; Østgaard et al., 2000, 2001, 2002a, b; Chua et al., 2001; Christensen et al., 2003;
Sætre et al., 2004]. Of particular interest is the technique used by Østgaard et al. [2001], combining UV and X-ray measurements to extend the valid energy range of the derived electron spectra. Such a combination of different remote sensing techniques allows a much better understanding of the auroral dynamics, as the various auroral forms are characterized by different precipitating electron energy spectra [Robinson and Vondrak, 1994]. The data base established by Østgaard et al. [2001, 2002a, b] have been used by Aksnes et al. [2002, 2004, 2005] to infer height profiles of and thereafter calculate the ionospheric conductances.
A few studies have examined the energy characteristics derived from UVI and PIXIE data.
The procedure used to calculate mean energies from UVI measurements was investigated by Germany et al. [2001], revealing an upper limit of 23 for modelling errors. Non-modelling errors like image processing and Poisson uncertainties have been assigned 3 and 5
respectively [Germany et al., 1997]. In a study by Østgaard et al. [2001], the electron energy fluxes calculated from combined UVI and PIXIE measurements were compared with in-situ particle measurements from the DMSP spacecraft. For the electron energy range between 0.09 and 30 keV, Østgaard et al. [2001] calculated an average ratio of 1.03 0.6 between the measured and derived energy flux. Despite relatively large differences for some of the data points, the results by Østgaard et al. [2001] indicate that the energy characteristics deduced from Polar satellite remote sensing data are reasonable. A complicating issue, though, is the large difference in resolution between the two techniques, which may explain the large standard deviation found by Østgaard et al. [2001]. In another comparison study by Aksnes et al.
[2005], the conductances calculated from UVI and PIXIE measurements were compared with similar conductances derived using a ground-based technique called Method of Characteristics [Inhester et al., 1992; Amm, 1995, 1998; Sillanp ¨a¨a, 2002]. A correlation coefficient of 0.57 was found between the two conductance sets. This number may seem low, but a number of complicating factors led Aksnes et al. [2005] to conclude that a fairly acceptable agreement had been reached.
The results obtained by Germany et al. [1997, 2001], Østgaard et al. [2001], and Aksnes et al. [2005] are promising, suggesting that the energy characteristics derived from the Polar satellite remote sensing data are reliable. We should be cautious with this conclusion, though, considering the limited statistics and the large discrepancies between some of the data points in the studies mentioned above. As already stated, the most reliable measurements of are probably provided by incoherent scatter radars. Doe et al. [1997] used UVI measurements to derive the auroral energy flux. The same quantity was then derived from
profiles measured
by the Sondrestrom radar. During a 16-min time period from an event on 20 May 1996, Doe et al. [1997] found that the total energy flux from the radar measurements agreed to within the
30 uncertainty of the UVI estimates.
In this study we compare Polar satellite imaging data with measurements from an incoherent scatter radar, similar to Doe et al. [1997]. Our approach is different, as we derive height profiles of from UV and X-ray emissions, using the procedures described by Østgaard et al. [2001] and Aksnes et al. [2002, 2004]. These derived
values are then compared with simultaneously measured from the European Incoherent Scatter (EISCAT) radar. In Sect. 2, the two techniques are presented, followed by results (Sect. 3) and discussion (Sect. 4). Finally, we conclude in Sect. 5.
2. Data and techniques
2.1. from EISCAT radar data
In this study we use data from the EISCAT UHF radar in Tromsø, Norway. The incoherent scatter radar technique adopted by EISCAT is considered the most proper method to derive [Robinson and Vondrak, 1994]. This procedure relies on the transmission of coded pulses of electromagnetic energy that in the ionosphere interact with free electrons through the process of Thomson scattering. Since all target electrons have individual thermal motion the total reflected signal will appear to be incoherent. By decoding the received signal the power spectrum or the auto-correlation function can be determined, and geophysical parameters like the electron density, the electron temperature, the ion temperature and the line-of-sight ion drift can be
derived from the shape of the spectrum. The EISCAT radars are capable of operating within the altitude range between 50 and 2500 km with a temporal resolution varying from less than a second to tens of minutes, and with spatial resolution less than 200 m. However, the EISCAT radars typically operate in a more specific mode that is optimized for the particular geophysical process to be studied.
In this study we have investigated a substorm event taking place on 24 March 1998, when the EISCAT UHF radar was running in a mode known as Common Programme Two (CP2).
Table 1.
During the CP2 operation mode the radar antenna shifts between four different beam directions, according to the specifications listed in Tab. 1. We note that the integration time varies between 50 and 80 s, giving a total cycle time of 6 min. Further, the
profiles have a spatial resolution of 1 km in the horizontal direction and 5 km in altitude.
2.2. from remote sensing of UV and X-ray emissions 2.2.1. The Ultraviolet Imager (UVI)
The UVI camera onboard Polar measures UV emissions, produced when precipitating particles excite the neutral atmosphere through collisions. A filter wheel allows the instrument to select between different UV spectral regions, including emissions within the Lyman-Birge- Hopfield (LBH) band (140-180 nm). By dividing these measurements into LBHS (140-160 nm) and LBHL (160-180 nm), the precipitating electron energy characteristics can be extracted. This is due to a varying amount of absorption by molecules within the LBH band, peaking at the shortest wavelengths and being much less at longer wavelengths. While the ratio between the
two wavelength bands is used to determine the mean electron energy , the energy flux is calculated from LBHL.
Using and derived from UVI measurements, we can provide the spectral distribution of the precipitating electrons assuming the following exponential form:
(1)
where is the differential electron flux, is a proportionality factor, is the electron energy and
is the characteristic electron energy. Alternatively a Maxwellian distribution may be used:
! #" $%& '" ()*
'"
(2)
It has been known that the UVI technique of estimating mean energy from the LBH ratio fails to work for precipitating electrons in the energy range of 1 keV or less. This is due to the lack of absorption in the upper atmosphere above 150 km [Germany et al., 1998a], yielding a nearly constant LBH ratio. The results of the present study provide insights on the applicability of UVI-derived energy estimates of mean energy and energy flux, suggesting that the lower energy threshold of 1 keV may be closer to 2 keV.
Germany et al. [2001] showed that for mean energies above 10 keV, assuming a nominal 1 erg aurora, almost all the LBHS intensity is absorbed (their Fig. 1), limiting the use of the LBHL/LBHS ratio for energies much higher than 10 keV.
On the other hand, energy flux estimates are made from LBHL only and are relatively insensitive to absorption. (Some shorter wavelength emissions are passed by LBHL, so there is a weak absorption effect). Consequently energy flux estimates should be valid from at
least 0.2 to 25 keV.
Nominally, the UVI data have a spatial resolution of 40 km [Torr et al., 1995], but wobbling of the despun platform on the Polar satellite has degraded this resolution in one direction to 360 km. For the UVI operating modes used in this study, we have a temporal resolution of almost 2 min. The reader is referred to work by Torr et al. [1995] and Germany et al. [1997, 1998a, b] for more details about the UVI camera and the technique used to derive precipitating electron energy spectra.
2.2.2. The Polar Ionospheric X-ray Imaging Experiment (PIXIE)
The interaction between precipitating electrons and the nuclei of atmospheric particles causes production of X-ray bremsstrahlung. Such X-ray emissions are measured by the PIXIE camera on Polar. The efficient PIXIE X-ray detection range is between 2 and 22 keV, allowing an estimation of the precipitating electron spectrum between 3 and 100 keV. In this process of obtaining energy characteristics from PIXIE X-ray data, a look-up table based on an electron-photon transport code (developed from neutron transport codes [Lorence, 1992]) is used. This look-up table provides values of the X-ray production emitted at different zenith angles for different exponential electron energy spectra. The technique developed (see Østgaard et al. [2000] for details) yields a four-parameter spectral distribution of the precipitating electrons.
A great advantage with the PIXIE X-ray technique is that the temporal resolution may be set freely from event to event, not being fixed in time. However, to derive a proper electron spectrum, a sufficient number of detected X-ray photons is needed. This yields an integration
time of 6 min used in this study, while the spatial resolution is 700 km (due to Polar altitude of 7 and the size of the PIXIE pinholes).
2.2.3. Deriving precipitating electron energy spectra from combined UVI and PIXIE measurements
As stated in Sects. 2.2.1 and 2.2.2, the electron energy characteristics extracted from UVI and PIXIE measurements correspond to different energy ranges. However, by combining these two remote sensing techniques, we can derive precipitating electron energy spectra valid from less than 1 keV to 100 keV. A procedure has been developed [Østgaard et al., 2001], in which UVI measurements are averaged within the actual PIXIE time frames and spatial resolution area.
The resulting electron energy parameters derived from UVI data are fitted to an exponential and a Maxwellian distribution (Eqs. 1-2). We then choose the UVI-derived spectrum that gives the smoothest transition to the PIXIE-derived spectrum.
2.2.4. Deriving using MANGLE
A number of studies have investigated the electron transport in the atmosphere [e.g. Gr ¨un, 1957; Rees, 1963; Mæhlum and Stadsnes, 1967; Walt et al., 1967; Banks and Nagy, 1970; Nagy and Banks, 1970; Banks et al., 1974; Strickland et al., 1976; Vondrak and Baron, 1976; Vondrak and Robinson, 1985; Solomon et al., 1988; Solomon, 1989, 1993; Rees and Lummerzheim, 1989;
Richards and Torr, 1990], varying in complexity and reliability. In this study we have used the MANGLE code to calculate the electron density caused by the precipitating electron energy spectra derived from UVI and PIXIE measurements. MANGLE is a relatively simple model,
developed by University of Maryland, based on the TANGLE code [Vondrak and Baron, 1976;
Vondrak and Robinson, 1985], and later modified by Aksnes [2005]. In the following, we will give a short description on how is calculated using MANGLE.
The ionospheric electron source term may be derived using the following equation:
(3)
where is the altitude, is the atmospheric depth,
is the atmospheric mass density, ! is the differential electron current flux, and" (= 35 eV) is the average energy needed to produce an ion-electron pair. is the electron energy deposition function from the cosine-dependent Isotropic over the Downward Hemisphere (IDH) model of Rees [1963]. is taken from work by Berger et al. [1974] and gives the ionization due to bremsstrahlung X-rays. For additional information about the various parameters listed above, see Aksnes [2005].
The electron source term from Eq. 3 is related to the electron density through the time-dependent rate equation:
" $#%'&
)(
+*,*
(4)
where is the background electron source term calculated using the International Reference Ionosphere (IRI)-95 model, and ( +*,* is the effective dissociative recombination coefficient taken from formulas by Vickrey et al. [1982] and Gledhill [1986]. Note that diffusive transport has not been included into Eq. 4, as characteristic time scales for diffusion in the E-region is on the order of hours and days [Richmond, 1995a]. Further, we may assume chemical equilibrium, as the recombination time is on the order of seconds around 100 km. This gives
$#-.& 0/
,
yielding the following expression of :
!
(
+*,* (5) 2.3. 24 March 1998
In this study we have derived and compared electron density profiles using the two
techniques described in Sects. 2.1-2.2, during a substorm event occurring on 24 March 1998. Figure 1.
The geomagnetic and indices between 2030 and 2330 UT are presented in Fig.
1. During this 3-hour time period, the geomagnetic disturbances suggest significant substorm activity. While drops to almost -700 nT ( 2115 UT), increases to 900 nT. This substorm activity takes place during the main phase of a moderate geomagnetic storm, as the
&
index drops from 0 nT to almost -50 nT within a 12-hour period (not shown here).
The numbers 1-18 and their corresponding vertical dashed lines in Fig. 1 indicate the times of the 18 plots presented in Figs. 2-3 (Sect. 3).
3. Results
As explained in Sect. 2.2.1, UV-emissions provide information about the lower electron energies up to 10-15 keV. To capture the high-energy tail of the precipitating electrons, we also need X-ray measurements (Sect. 2.2.2). While this suggests that both data sets should be included in the comparison analysis, the coarse PIXIE resolution severely complicates a comparison with EISCAT measurements. We have therefore established two different procedures in this study when examining electron density values derived from Polar satellite imaging data. The first approach includes UV-emissions only, yielding profiles valid in
the E-region above 105 km (Sect. 3.1). Then we have included PIXIE measurements (Sect.
3.2) , resulting in values in the whole E-region and upper D-region.
3.1. vs
The first investigation is a comparison between electron density values derived from UVI measurements, , with similar!" profiles using EISCAT radar data. The values of mean energy and energy flux derived from UVI are represented using an exponential and a Maxwellian distribution. Then we have chosen, in each case, the spectrum which yields the best match between and . The altitude region of interest goes from 105 to 155 km.
The comparison is complicated by the differences in temporal resolution. It takes about 2 min to perform an energy analysis for the UVI operating mode in this study, while the
profiles have an integration time varying between 50 and 80 s (Tab. 1). We have compared each set of values with the profile closest in time.
Another difficulty arises from the difference in spatial resolution. While UVI cannot resolve structures less than 40 km (Sect. 2.2.1), EISCAT gives profiles in the range of 1 km (Sect. 2.1). As EISCAT looks in four different directions during the CP2 mode on 24 March 1998 (Tab. 1), the EISCAT measurements cover a geographic region of 68.8-69.6 in latitude and 19.2-21.2 in longitude. The profiles used in this study are taken from a larger region surrounding the EISCAT beam locations and corresponding to an area of 100 km x 150 km. This larger region, exceeding the UVI nominal resolution of 40 km, is chosen to get sufficient count rates to derive reliable electron energy characteristics.
The difference in resolution suggests that discrepancies are expected to occur when we have discrete precipitation within the UVI field of view (FOV). To minimize such effects, we have performed an investigation of the individual UVI pixel intensities within the UVI region of
100 km x 150 km. Only cases where all the individual UVI pixel intensities are within 50 of the UVI mean value are included in the comparison analysis. The procedure performed is similar to the one described in Aksnes et al. [2005], meaning that we only include cases with
relatively uniform precipitation within the UVI analysis area. Figure 2.
In Fig. 2 we give height profiles of the electron density derived using UVI-measurements (solid lines) and EISCAT data (horizontal bars), during the 18 time periods when the particle precipitation is relatively uniform in a larger region surrounding the EISCAT FOV. The length of the horizontal bars gives the !" uncertainties, while the dashed lines represent the uncertainties. The latter have been calculated using the standard deviation of the UVI mean energy and the energy flux. A value is presented in each plot, giving the percentage number of EISCAT data points that fall within the UVI error limits. From Fig. 2 we see that some of the data sets reveal a pretty good match between the two techniques. E.g. in plots no.
16-18 we find that all the
!"
values are within the uncertainties (100 ).
Other plots reveal significant discrepancies between the two techniques. In particular, the profiles deviate strongly in cases where shows a maximum in the upper E-region.
The horizontal solid line in each plot gives the altitude of the
maximum. Usually we find the largest!" values below 110 km. However, in plots no. 3, 5, 6, 10, and 12, the maximum in!" takes place above 130 km (indicated by the horizontal dashed line).
As pointed out in Sect. 2.2.1, the lack of in the upper atmosphere yields an almost constant
LBHL/LBHS ratio, meaning that the UVI technique may fail to work. For these five cases in Fig. 2 with!" maxima above 130 km, the values representing the match between the two techniques vary from 12 to 50 . In comparison, the remaining 13 plots have corresponding values of match between 55 and 100 .
We have recalculated for the cases when EISCAT data indicate a dominance of lower electron energies, using the following approach: First we determine the characteristic energy
(Eqs. 1-2) from the peak value of
. Note that this procedure is complicated by the relatively constant values as a function of altitude for plots no. 5, 6, 10, and 12. This suggests that the chosen peak values, determined by identifying the altitudes with the largest values, are associated with a significant degree of uncertainty. The next step in our procedure is to modify the corresponding value of , keeping the UVI electron energy flux fixed at the same value as earlier (derived from the UVI-LBHL measurements). may be expressed:
$
(6)
where is taken from Eqs. 1-2. Figure 3.
In Fig. 3 we have replotted the five plots from Fig. 2 when the values reveal their maxima in the upper E-region above 130 km, using the modified height profiles of . This new approach clearly gives a much better agreement between the two techniques, showing that UVI measures about the same energy flux as EISCAT in the upper E-region. From Fig. 3, we note that the number of points with match between and !" exceed 70 in three of the plots.
3.2. vs
As demonstrated in Aksnes et al. [2002, 2004], the PIXIE X-ray data are needed to capture the high-energy tail of the precipitating electrons, depositing their energy in the lower E-region (below 105 km) and upper D-region. Such X-ray data are available between 2100 and 2125 UT on 24 March 1998. Within this time period, we have derived electron spectra from PIXIE using a time resolution of 6 min. All available UVI energy parameters within the PIXIE time intervals have been averaged, yielding
profiles valid between 75 and 155 km altitude. Also, note that the given time period of 2100-2125 UT involves the modified UVI
plots no. 3, 5, and 6 from Fig. 3. Figure 4.
In Fig. 4 the four
profiles (solid thin line) are compared with similar 6-min averaged values (solid thick line). The PIXIE pixel size of 700 km is too large to allow an accurate comparison with EISCAT measurements. Still, we find that the two techniques give results that match fairly well. As shown in Fig. 4, most of the
data points (78 to 86 ) are within the dashed lines indicating the uncertainties of the satellite remote sensing measurements.
4. Discussion
Space-based remote sensing of UV and X-ray emissions provide a powerful tool to investigate the ionospheric electrodynamics. In this study we have performed a ground-truth experiment of the electron density values derived from UVI and PIXIE measurements, by comparing with simultaneously measured
using EISCAT radar data.
To reduce errors caused by the difference in spatial resolution between the two techniques, we have excluded cases with discrete precipitation. As shown in Fig. 2, many of the plots reveal a good agreement between and !" . Still, large discrepancies may sometimes be observed. This is to be expected, as finer structures in the auroral dynamics will be effectively smoothed out in the satellite imaging data. Therefore randomly distributed deviations should occur from time to time. However, the results presented in Fig. 2 also reveal a systematic dominance of
over in the upper E-region during times when the maximum in takes place above 130 km. Ionization maxima at these high altitudes suggest that large parts of the electron precipitation are soft particles of energy 2 keV or less [Rees, 1963]. Such conditions are not favorable when extracting the energy characteristics from UVI measurements. The reason is that the UV-technique relies strongly on absorption of UV-emissions, which drops significantly at higher altitudes. We therefore suggest that the discrepancies revealed in plots no. 3, 5, 6, 10, and 12 in Fig. 2 is caused by inaccurate
values derived from UVI data. A special approach is performed for these five cases, where height of the peak value of yields a new value of the characteristic energy for the UVI spectrum. (Eqs. 1-2). We then adjust
accordingly, given that the UVI energy flux derived from UVI-LBHL measurements is correct. The results found show that UVI is measuring more or less the same energy flux as EISCAT.
The inclusion of PIXIE X-ray measurements in Fig. 4 allows us to study a wider altitude range of 75-155 km. The discrepancy in spatial resolution means we must be cautious when interpreting the results, as the PIXIE pixel size is too large to make an accurate comparison with EISCAT data. Still, we find a good match in all four plots, supporting the technique used to
derive from satellite remote sensing measurements. Fig. 4c (between 2113 and 2119 UT)
actually reveals almost identical profiles in the lower E-region and upper D-region. Figure 5.
In Fig. 5 we have calculated the Pedersen conductances, , using the 18 (+) and
( ) profiles presented in Fig. 2. A fairly good match is revealed in both trend and magnitude. The pluses (+) in the bottom panel of Fig. 5 indicate how much the Pedersen conductances calculated using UVI measurements, , differ from !" . For 15 of the 18 cases, the two sets of values are within 30 . Also, we find that despite the large altitude variation (Figs. 2-3) the 5 modified profiles yield Pedersen conductance values (
) similar to the original profiles (as they almost cover the corresponding +’es). This finding suggests that UV-emissions may be a reliable tool to monitor the Pedersen conductance even during periods when soft precipitation complicates the estimation of the mean electron energy.
The UV-technique allows only a two-parameter spectral distribution of the precipitating electrons to be determined. This can explain the apparent underestimation of values at lower heights in Fig. 3, where the Hall conductance is significant. Earlier studies by Aksnes et al. [2002, 2004] have concluded that X-ray data are needed to derive accurate values of the Hall conductance, as the X-ray photons can be used to capture the high-energy tail of the precipitating electrons. This is most important, because the precipitating electron energy
spectrum often changes shape and flatten out at higher energies. Figure 6.
The Hall and Pedersen conductances calculated using the four (solid lines) and (dashed lines) profiles from Fig. 4, are presented in Fig. 6, revealing a fairly good match between the two conductance sets. While
is within 20 of
!"
, we find that is 2 to 26 larger than .
From the results found in this study, we may argue that remote sensing of UV and X-ray emissions provide a powerful tool to investigate the ionospheric electrodynamics. Still, Figs.
4 and 6 indicate that
derived from space-based remote sensing is slightly larger than
. A similar tendency is seen in Figs. 2 and 5.
Table 2.
By calculating the total height-integrated electron density
using the profiles presented in Figs. 2, 3 and 4, we find a pretty good match between the data sets. However, a closer look at the data reveal that
and
are usually slightly larger than
. The 18
/
ratios presented in Tab. 2 yield an average value of 1.05, while the 4
/
ratios give a similar value of 1.13. This moderate overestimation may be due to the large differences in spatial resolution, causing gradients in the precipitation to be smoothed when using the remote sensing technique. Even though we include only cases where the precipitation is relatively uniform, the UVI (PIXIE) resolution of 40 km ( 700 km) means that structures of less spatial extent cannot be revealed. Instead such structures are smoothed. This may result in an overestimation of the electron density. Assume we divide a region of interest in subregions. In situation 1, we have a very localized precipitation so that the total electron production rate
takes place within one of the subregions. For this subregion, . As / in the other subregions, the averaged over the whole region is:
(
(7) In situation 2, we now assume that the same total electron production rate as in situation 1
takes places homogeneously over the whole region, meaning that the electron production rate is
"#
within each of the subregions. This results in a larger :
(
(8)
Note that the same arguments also goes for differences in temporal resolution. We should add that for small auroral features and large field-of-view per pixel, a number of complicating issues may exist. The problems with differences in spatial resolution, as described above, is only one piece of the puzzle. However, a properly evaluation of these issues is beyond the scope of this paper.
As shown in this study, LBH-measurements provide fairly reliable electron density profiles for most of the E-region, given mean electron energies between 2 and 10 keV. During conditions with a very soft precipitation present, though, other remote sensing techniques are needed. In Strickland et al. [1983], the intensity of selected visible and FUV emissions as a function of the Maxwellian characteristic energy (their Fig. 7) is given. While the 127.3 nm emissions are almost unattenuated regardless of the altitude at which the energy is deposited, the 135.6 nm emissions reveal a strong dependence with mean energy. This suggests that simultaneous measurements of 127.3 and 135.6 nm emissions could be used to infer the low-energy electrons below 2 keV. In Robinson and Vondrak [1994], they show that the ratio between 127.3 and 135.6 nm emissions is proportional with the mean electron energy from a few hundred eV to more than 10 keV. Another approach is proposed by Semeter et al. [2001], suggesting to use two spectral bands in the visible region (centered at 427.8 and 732.5 nm) to characterize the precipitating electrons below 1 keV. The remote sensing technique relies
on steady-state conditions when solving the continuity equation (Eq. 5), and at high altitudes (above 150 km) the electron density may not reach its steady-state values. Consequently, we may experience differences between derived from satellite data and , even though the energy characteristics extracted from the satellite measurements are indeed correct.
5. Conclusion
In this study we have performed a ground-truthing of derived from satellite imaging measurements of UV and X-ray emissions, by comparing height profiles of and
with
values using the EISCAT UHF radar in Tromsø, Norway. For a substorm event occurring on 24 March 1998, we have investigated 18 and !"
profiles between 105 and 155 km from time periods when the particle precipitation is relatively uniform in a larger region surrounding the EISCAT radar facility. Many of the profiles reveal a good match, supporting the validity of the space-based technique. The largest discrepancies occur during time periods when the !" values indicate that a significant portion of precipitating electrons is in the energy range of 2 keV or less. We argue that the derived
profiles are not valid during such conditions, as the UV-emissions produced in high altitudes are hardly absorbed by . Consequently, the UV-technique used to derive an auroral average electron energy fails to work. A recalculation of the profiles using information of the height of the maximum in from EISCAT radar measurements reveals that UVI is measuring more or less the same energy flux as EISCAT. Despite the difficulties in obtaining a proper height profile of during time periods with soft precipitation, the effects on the Pedersen conductances turn out to be insignificant. A comparison with EISCAT calculations
further shows that are within 30 of for 15 of 18 cases, indicating that UV-emissions are a powerful tool to examine the Pedersen conductance.
The inclusion of PIXIE X-ray data in the comparison analysis has extended the satellite- derived
profiles to include the lower E-region and upper D-region. The analysis performed cannot be considered an absolute validation of the value of the combined UV/X-ray method.
This is due to the large PIXIE pixel size, preventing an accurate comparison with EISCAT data.
The technique presented should work fine, though, if X-ray measurements with a much higher spatial resolution were available. Still, the results are promising, showing relatively good match between!" and . We further find a pretty good match (within 25 ) when calculating the Hall and Pedersen conductances using the two techniques.
An investigation of the total electron density values
reveals that the satellite imaging measurements give slightly larger values than the ground-based EISCAT data. This may be explained by the difference in resolution, as smoothing of discrete precipitation should yield a larger value of . While
is 5 larger than
!"
, this number increases to 13
when including PIXIE X-ray measurements. Figure 7.
To summarize our result, we show in Fig. 7 the validity of remote sensing techniques as a function of altitude of peak energy deposition. During periods with a mean electron energy
between 2 and 10 keV, corresponding to peak energy deposition within an altitude region of
105-130 km [Rees, 1963], the UVI technique yields a fairly good determination of the electron density in the whole E-region down to 100 km. At lower altitudes, PIXIE data are needed to capture the high-energy tail of the precipitating electrons. For those situations with very soft precipitation, additional information from other remote sensing techniques that work for
below 2 keV is needed.
Acknowledgments. This study was supported by the Research Council of Norway (NFR).
EISCAT is an international association supported by Finland (SA), France (CNRS), Germany (MPG), Japan (NIPR), Norway (NFR), Sweden (NFR), and the United Kingdom (PPARC). This work was supported by NASA UVI funding from University of California Berkeley contract SA3527 to the University of Alabama in Huntsville.
References
Aksnes, A., Ionospheric global conductances derived from remote sensing of UV and X-ray emissions, Ph.D. thesis, Univ. of Bergen, Bergen, Norway, 2005.
Aksnes, A., J. Stadsnes, J. Bjordal, N. Østgaard, R. R. Vondrak, D. L. Detrick, T. J. Rosenberg, G. A. Germany, and D. Chenette, Instantaneous ionospheric global conductance maps during an isolated substorm, Ann. Geophys., 20, 1181, 2002.
Aksnes, A., J. Stadsnes, G. Lu, N. Østgaard, R. R. Vondrak, D. L. Detrick, T. J. Rosenberg, G. A. Germany, and M. Schulz, Effects of energetic electrons on the electrodynamics in the ionosphere, Ann. Geophys., 22, 475, 2004.
Aksnes, A., O. Amm, J. Stadsnes, N. Østgaard, G. A. Germany, R. R. Vondrak, and I. Sillanp¨a¨a, Ionospheric conductances derived from satellite measurements of auroral UV and X-ray emissions, and ground-based electromagnetic data: A comparison, Ann. Geophys., 23, 343, 2005.
Amm, O., Direct determination of the local ionospheric Hall conductance distribution from two-dimensional electric and magnetic field data: application of the method using models of typical ionospheric electrodynamic situations, J. Geophys. Res., 100, 24,173, 1995.
Amm, O., Method of characteristics in spherical geometry applied to a Harang-discontinuity situation, Ann. Geophys., 16, 413, 1998.
Banks, P. M., and A. F. Nagy, Concerning the influence of elastic scattering upon photoelectron
transport and escape, J. Geophys. Res., 75, 1902, 1970.
Banks, P. M., C. R. Chappell, and A. F. Nagy, A new model for the interaction of auroral electrons with the atmosphere: Spectral degradation, backscatter, optical emission, and ionization, J. Geophys. Res., 79, 1459, 1974.
Berger, M. J., S. M. Seltzer, and K. Maeda, Some new results on electron transport in the atmosphere, J. Atmos. Terr. Phys., 36, 591, 1974.
Brittnacher, M., R. Elsen, G. Parks, L. Chen, G. Germany, and J. Spann, A dayside auroral energy deposition case study using the Polar Ultraviolet Imager, Geophys. Res. Lett., 24, 991, 1997.
Christensen, T., N. Østgaard, T. J. Rosenberg, D. L. Detrick, G. A. Germany, and P. Stauning, Conjugate high-intensity energetic electron precipitation at high latitude, Ann. Geophys., 21, 1443, 2003.
Chua, D., G. Parks, M. Brittnacher, W. Peria, G. Germany, J. Spann, and C. Carlson, Energy characteristics of auroral electron precipitation: A comparison of substorms and pressure pulse related auroral activity, J. Geophys. Res., 106, 10.1029/2001JA003,027, 2001.
Doe, R. A., J. D. Kelly, D. Lummerzheim, G. K. Parks, M. J. Brittnacher, G. A. Germany, and J. Spann, Initial comparison of POLAR UVI and Sondrestrom IS radar estimates for auroral electron energy flux, Geophys. Res. Lett., 24, 999, 1997.
Germany, G. A., G. K. Parks, M. Brittnacher, J. Cumnock, D. Lummerzheim, J. F. Spann, L. Chen, P. G. Richards, and F. J. Rich, Remote determination of auroral energy characteristics during substorm activity, Geophys. Res. Lett., 24, 995, 1997.
Germany, G. A., G. K. Parks, M. Brittnacher, J. F. Spann, J. Cumnock, D. Lummerzheim, F. J.
Rich, and P. G. Richards, Energy characterization of a dynamic auroral event using GGS UVI images, in Geospace Mass and Energy Flow: Results from the International Solar-Terrestrial Physics Program, edited by J. L. Horwitz, D. L. Gallagher, and W. K.
Peterson, Geophysical Monograph Series, p. 143, AGU, 104, Washington, D. C., USA, 1998a.
Germany, G. A., J. F. Spann, G. K. Parks, M. Brittnacher, R. Elsen, L. Chen, D. Lummerzheim, and M. Rees, Auroral observations from the Polar Ultraviolet Imager (UVI), in Geospace Mass and Energy Flow: Results from the International Solar-Terrestrial Physics Program, edited by J. L. Horwitz, D. L. Gallagher, and W. K. Peterson, Geophysical Monograph Series, p. 149, AGU, 104, Washington, D. C., USA, 1998b.
Germany, G. A., D. Lummerzheim, and P. G. Richards, Impact of model differences on quantitative analysis of FUV auroral emissions: Total ionization cross sections, J.
Geophys. Res., 106, 12,837, 2001.
Gledhill, J. A., The effective recombination coefficient of electrons in the ionosphere between 50 and 150 km, Radio Sci., 21, 339, 1986.
Gr¨un, A. E., Lumineszenz-photometrische Messungen der Energie-Absorption im Strahlungs- feld von Electronquellen, Eindimensionaler Fall in Luft, Z. Naturforsch., Ser. A, 12, 89, 1957.
Imhof, W. L., et al., The Polar Ionospheric X-ray Imaging Experiment (PIXIE), Space Sci. Rev., 71, 385, 1995.
Inhester, B., J. Untiedt, M. Segatz, and M. K¨urschner, Direct determination of the local ionospheric Hall conductance distribution from two-dimensional electric and magnetic field data, J. Geophys. Res., 97, 4073, 1992.
Lorence, L. J., CEPXS/ONELD Version 2.0: A discrete ordinates code package for general one-dimensional coupled electron-photon transport, IEES Trans. Nucl. Sci., 39, 1031, 1992.
Lummerzheim, D., M. Brittnacher, D. Evans, G. A. Germany, G. K. Parks, M. H. Rees, and J. F. Spann, High time resolution study of the hemispheric power carried by energetic electrons into the ionosphere during the May 19/20, 1996 auroral activity, Geophys. Res.
Lett., 24, 987, 1997.
Mæhlum, B., and J. Stadsnes, Scattering and absorption of fast electrons in the upper atmosphere, Physica Norvegica, 2, 111, 1967.
Nagy, A. F., and P. M. Banks, Photoelectron fluxes in the ionosphere, J. Geophys. Res., 75, 6260, 1970.
Østgaard, N., J. Stadsnes, J. Bjordal, R. R. Vondrak, S. A. Cummer, D. Chenette, M. Schulz, and J. Pronko, Cause of the localized maximum of X-ray emission in the morning sector: A comparison with electron measurements, J. Geophys. Res., 105, 20,869, 2000.
Østgaard, N., J. Stadsnes, J. Bjordal, G. A. Germany, R. R. Vondrak, G. K. Parks, S. A. Cummer, D. Chenette, and J. Pronko, Auroral electron distributions derived from combined UV and X-ray emissions, J. Geophys. Res., 106, 26,081, 2001.
Østgaard, N., G. A. Germany, J. Stadsnes, and R. R. Vondrak, Energy analysis of substorms
based on remote sensing techniques, solar wind measurements and geomagnetic indices, J. Geophys. Res., 107, 1233, doi: 10.1029/2001JA002,002, 2002a.
Østgaard, N., R. R. Vondrak, J. M. Gjerloev, and G. A. Germany, A relation between the energy deposition by electron precipitation and geomagnetic indices during substorms, J. Geophys. Res., 107, 1246, doi: 10.1029/2001JA002,003, 2002b.
Rees, M. H., Auroral ionization and excitation by incident energetic electrons, Planet. Space Sci., 11, 1209, 1963.
Rees, M. H., and D. Lummerzheim, Characteristics of auroral electron precipitation derived from optical spectroscopy, J. Geophys. Res., 94, 6799, 1989.
Richards, P. G., and D. G. Torr, Auroral modeling of the 3371 ˚ emission rate: Dependence on characteristic electron energy, J. Geophys. Res., 95, 10,337, 1990.
Richmond, A. D., Ionospheric electrodynamics, in Handbook of Atmospheric Electrodynamics, Vol. II, edited by H. Volland, p. 249, CRC Press, Boca Raton, Florida, USA, 1995a.
Robinson, R. M., and R. R. Vondrak, Validation of techniques for space based remote sensing of auroral precipitation and its ionospheric effects, Space Sci. Rev., 69, 331, 1994.
Robinson, R. M., R. R. Vondrak, J. D. Craven, L. A. Frank, and K. Miller, A comparison of ionospheric electron densities and auroral luminosities observed simultaneously with the Chatanika radar and the DE-1 auroral imagers, J. Geophys. Res., 94, 5382, 1989.
Robinson, R. M., et al., Coordinated measurements made by the Sondrestrom radar and the Polar Bear ultraviolet imager, J. Geophys. Res., 97, 2863, 1992.
Sætre, C., J. Stadsnes, H. Nesse, A. Aksnes, S. M. Petrinec, C. A. Barth, D. N. Baker, R. R.
Vondrak, and N. Østgaard, Energetic electron precipitation and the NO abundance in the upper atmosphere: A direct comparison during a geomagnetic storm, J. Geophys. Res., 109, 10.1029/2004/JA010,485, 2004.
Semeter, J., D. Lummerzheim, and G. Haerendel, Simultaneous multispectral imaging of the discrete aurora, J. Atmos. Terr. Phys., 63, 1981, 2001.
Sillanp¨a¨a, I., One-dimensional method of characteristics to determine ionospheric conductances and currents, Ph.D. thesis, Helsinki, Finland, 2002.
Solomon, S. C., Auroral excitation of the 2P(0,0) and VK(0,9) bands, J. Geophys. Res., 94, 17,215, 1989.
Solomon, S. C., Auroral electron transport using the Monte Carlo method, Geophys. Res. Lett., 20, 185, 1993.
Solomon, S. C., P. B. Hays, and V. J. Abreu, The auroral 6300 ˚ emission: Observations and modeling, J. Geophys. Res., 93, 9867, 1988.
Strickland, D. J., D. L. Book, T. P. Coffey, and J. A. Fedder, Transport equation techniques for the deposition of auroral electrons, J. Geophys. Res., 81, 2755, 1976.
Strickland, D. J., J. R. Jasperse, and J. A. Whalen, Dependence of auroral FUV emissions on the incident electron spectrum and neutral atmosphere, J. Geophys. Res., 88, 8051, 1983.
Torr, M. R., et al., A far ultraviolet imager for the international solar-terrestrial physics mission, Space Sci. Rev., 71, 329, 1995.
Vickrey, J. F., R. R. Vondrak, and S. J. Matthews, Energy deposition by energetic particles and Joule dissipation in the auroral ionosphere, J. Geophys. Res., 87, 5184, 1982.
Vondrak, R. R., and M. J. Baron, Radar measurements of the latitudinal variation of auroral ionization, Radio Sci., 11, 939, 1976.
Vondrak, R. R., and R. Robinson, Inference of high-latitude ionization and conductivity from AE-C measurements of auroral electron fluxes, J. Geophys. Res., 90, 7505, 1985.
Vondrak, R. R., R. M. Robinson, P. F. Mizera, and D. J. Gorney, X-ray spectrophotometric remote sensing of diffuse auroral ionization, Radio Sci., 23, 537, 1988.
Walt, M., W. M. MacDonald, and W. E. Francis, Penetration of auroral electrons into the atmosphere, in Physics of the Magnetosphere, edited by R. L. Carovillano, J. F. McClays, and H. R. Radoski, p. 534, D. Reidel, Dordrecht, Netherlands, 1967.
A. Aksnes, Florida Space Institute, MS-FSI, Kennedy Space Center, FL 32899, U.S.A.
J. Stadsnes, N. Østgaard, University of Bergen, Department of Physics and Technology, Allegt. 55, 5007 Bergen, Norway ([email protected];[email protected])
G. A. Germany, University of Alabama in Huntsville, AL 35899, U.S.A. (ger- [email protected])
K. Oksavik, Johns Hopkins University/Applied Physics Laboratory, 11100 Johns Hopkins Road, Laurel, MD 20723, U.S.A. ([email protected])
R. R. Vondrak, NASA/Goddard Space Flight Center, Greenbelt, MD 20771, U.S.A.
A. Brekke, U. P. Løvhaug, University of Tromsø, Department of Physics, 9037 Tromsø, Norway ([email protected];[email protected])
Received
Submitted for publication in Journal of Geophysical Research, June 2005
This manuscript was prepared with AGU’s LATEX macros v5, with the extension package
‘AGU ’ by P. W. Daly, version 1.6b from 1999/08/19.
Figure Captions
Figure 1. The and indices between 2030 and 2330 UT on 24 March 1998. The vertical dashed lines and their respective numbers ranging from 1 to 18 refer to the different times of the plots presented in Figs. 2-3.
105 130
155 (1)
70 %
(2) 75 %
Altitude [km]
(3) 30 %
(4) 55 %
105 130
155 (5)
40 %
(6) 12 %
Altitude [km]
(7) 80 %
(8) 100 %
1010 1011 1012
Ne [electrons per m3] 105
130
155 (9)
55 %
(10) 40 %
(11) 71 %
(12) 50 %
(13) 90 %
(14) 75 %
(15) 80 %
(16) 100 %
(17) 100 %
1010 1011 1012
Ne [electrons per m3] (18) 100 %
Figure 2. The electron density derived using UVI measurements (solid line) and EISCAT data (horizontal bars) on 24 March 1998. The dashed lines represent the uncertainty of the satellite measurements, and the percentages give the match between the two techniques. The horizontal solid line in each plot indicate the altitude of the maximum, while the horizontal dashed line is plotted at 130 km. The time periods for the plots 1-18 are indicated in Fig. 1.
105 130 155
(3) 70 %
(5) 30 %
(6) 75 %
(10) 30 %
(12) 75 %
(3) 70 %
(5) 30 %
(6) 75 %
(10) 30 %
(12) 75 %
Altitude [km]
(3) 70 %
(5) 30 %
(6) 75 %
(10) 30 %
(12) 75 %
(3) 70 %
(5) 30 %
(6) 75 %
(10) 30 %
(12) 75 %
1010 1011 1012
Ne [electrons per m3 ] 105
130 155
(3) 70 %
(5) 30 %
(6) 75 %
(10) 30 %
(12) 75 %
Figure 3. Modified profiles of (solid line) for the five plots in Fig. 2 where the
!"
values (horizontal bars) reveal maxima above 130 km. The dashed lines represent the uncertainty of the satellite measurements.
75 95 115 135 155
2101-2107 UT
2107-2113 UT
Altitude [km]
2113-2119 UT
1010 1011 1012
Ne [electrons per m3] 2119-2125 UT
(a)
(b)
(c)
(d) 83 %
86 %
85 %
78 %
Figure 4. The electron density derived using UVI and PIXIE measurements (solid thin line) and EISCAT data (solid thick line) on 24 March 1998. The dashed and dotted lines indicate the uncertainties of the satellite and radar measurements, respectively.
20:30 21:30 22:30 23:30 Time [UT]
-50 -25 0 25 50
[%]
20:30 21:30 22:30 23:30
0 5 10 15
Pedersen Conductance [S]
Figure 5. The Pedersen conductances (upper panel) calculated using the 18 (+) and
!"
( ) profiles from Fig. 2. Also plotted are the 5 modified (
) data sets from Fig. 3. The deviations between the two techniques ( ) are given in the bottom panel.
21:00 21:10 21:20 21:30 Time [UT]
[%]
-30 0 30
0 5 10 15
Pedersen Conductances [S]
21:00 21:10 21:20 21:30
Time [UT]
[%]
-30 0 30
0 10 20 30
Hall Conductances [S]
Figure 6. The Hall (upper panel) and Pedersen (bottom panel) conductances calculated using the four (solid lines) and (dashed lines) profiles from Fig. 4. Also plotted are the deviations ( ) between the two techniques.
75 85 95 105 115 125 135 145 155
Altitude [km]
75 85 95 105 115 125 135 145 155
Altitude [km] Mean Electron Energy
Remote sensing techniques that work in the low-energy domain
UVI
UVI and PIXIE
PIXIE
2 keV
5 keV
10 keV 15 keV
100 keV 200 keV
Figure 7. The validity of UVI and PIXIE measurements to determine the electron precipitation as a function of altitude of peak energy deposition. Left panel: The UVI technique works for mean electron energies between 2 and 10 keV, while PIXIE data captures the high-energy tail of the precipitating electrons. A transition altitude region exists for heights between 105 and 115 km, where both techniques work properly. Other remote sensing techniques are needed for conditions with below 2 keV. Right panel: The derived electron spectra from UVI are valid in the upper E-region above 100 km, while the PIXIE spectra are valid in the upper D-region and lower E-region, i.e. below 125 km.
Tables
Beam Direction Azimuth Elevation Time Resolution
1 133.3 54.5 65 s
2 183.2 77.2 50 s
3 180.0 89.9 80 s
4 166.5 57.2 65 s
Table 1.: EISCAT UHF radar beam directions for the event on 24 March 1998.
Azimuth angles of 0 and 90 correspond to geographic north and east respec- tively. The elevation angle is measured from the horizontal plane and upward.
Fig. 2:
#
1.19
1.08
1.15
1.17
1.06
1.35
1.40
1.18
1.16
1.07
0.89
0.99
0.94 Fig. 3:
#
0.79
0.65
0.84
1.13
0.92 Fig. 4:
#
1.17
1.16
1.00
1.19
Table 2.: Ratios between total height-integrated electron den- sity values, using the profiles presented in Figs. 2-4.
